Which measurements
can determine native action?

The anchor exponents must determine action.

Write each independently mapped anchor as y_j=k_jB₀^pⱼa₀^qⱼτ₀^rⱼ. Collect the exponents into W. The native action A₀=B₀a₀²/τ₀ is identifiable exactly when (1,2,−1) lies in the row span of W.

If the condition fails, an explicit positive rescaling preserves every anchor while changing action. If it passes, a row-span certificate gives a direct product estimator. Full rank three is sufficient but unnecessary.

Energy and time suffice under the declared map.

The physical apparatus-to-native map must be independently supported. A circuit’s measured energy times time cannot simply be relabeled as native action.

A scoped typed contract rejects the source defect.

Eighteen transcribed equalities pass dimensional checks across 17 bound source rows. The original O13 chain is rejected; the proposed separated reciprocal and action identities pass. The other 706 ledger rows remain outside this contract.

Negative controls reject energy-plus-action addition, a dimensional sine argument and an energy-typed B₀. Exact rational checks cover 2,952 anchor banks and their ambiguity witnesses.

Freeze the map before scoring the prediction.

For a fixed certificate c, bounded log-anchor errors propagate as |log(Â₀/A₀)|≤Σ|c_j|ε_j. This deterministic statement does not establish measurement coverage.

The next experimental return must supply the same-carrier map, coefficients, exponent matrix, target-blind calibration provenance, uncertainty bounds and frozen decision rule. Non-constructor review precedes scoring. QPB1’s η_A=2/1 remains quarantined; no experiment or physical promotion occurs here.